1

Vectorized Calculation of Short Circuit CurrentsConsidering Distributed Generation - An Open

Source Implementation of IEC 60909Leon Thurner, Martin Braun

Abstract—An important task in grid planning is to ensurethat faults in the grid are detected and cut off without dam-age in any grid element. Calculating short circuit currents istherefore a vital grid analysis functionality for grid planningapplications. The standard IEC 60909 provides guidelines forshort circuit calculations and is routinely applied in grid planningapplications. This paper presents a method for the vectorizedcalculation of short circuit currents according to IEC 60909.Distributed generation units are considered according to thelatest revision of the standard. The method is implemented inthe python based open source tool pandapower and validatedagainst commercial software and examples from literature. Theimplementation presented in this paper is the first comprehensiveimplementation of the IEC 60909 standard which is availableunder an open source license. It can be used to evaluate faultcurrents in grid studies with a high degree of automation and isshown to scale well for large grids. Its practical applicability isshown in a case study with a real MV grid with a high degreeof DG penetration.

Index Terms—short circuit calculation, fault current, IEC60909, VDE 0102, equivalent voltage source, distributed genera-tion, python, open source, pandapower .

I. INTRODUCTION

FAULTS in electric grids occur due to external interfer-ence, such as a tree falling on an overhead line, or due

to failures in grid elements, such as a disruptive discharge ina cable due to a failure in the insulation material. These faultslead to transient short circuit (SC) currents that can be higherthan normal operational currents by several magnitudes. Thesecurrents impose a high thermal as well as mechanical strainon lines, transformers and other power system components.Electrical grids have to be designed so that SC currents arelimited and faults can not lead to melting, combustion or evenexplosion in any grid components. It also has to be ensuredthat all faults can be detected and cut off by the protectionsystems. Calculating SC currents is therefore a crucial analysisfunctionality in grid planning and protection system design.

A. IEC 60909

Faults lead to transient currents that can be simulated withdynamic methods. In grid planning however, many possiblefault scenarios have to be considered in advance, which

L. Thurner is with the Department of Department of Energy Managementand Power System Operation, University of Kassel, 34121, Germany, e-mail: leon.thurner@uni-kassel.de M. Braun is with the Department of EnergyManagement and Power System Operation, University of Kassel and theFraunhofer IEE, 34121 Kassel, Germany

makes detailed dynamic simulations of each fault infeasible.To allow approximations of SC currents with static methods,the Technical Committee 73 of the IEC has put forwardthe standard IEC 60909 [1]. The standard characterizes theSC curve with an initial SC current I

′′

k , peak short circuitcurrent ip or long term SC current Ik as shown in Figure 1.Detailed instructions for how these currents can be calculatedfrom the static grid model are provided in the standard. Toaccount for transient effects as best as possible with staticcalculation methods, the standard defines multiple correctionfactors and other calculation rules. The most recent revisionof the standard also defines a method on how to integrate thecontribution of distributed generation (DG) to the SC current,which have previously been neglected [2]. The IEC 60909standard is widely applied in grid planning applications [3],[4], [5], [6], [7].

B. Available Tools

SC calculations according to IEC 60909 is a standardfunctionality in commercial power system analysis tools likeNeplan1, PSS Sincal2 or DIgSILENT PowerFactory3. Thesetools can be used by grid planners and are mostly designedas graphical user interface (GUI) applications where singlecalculations are triggered manually. For scientific purposeshowever, it is often necessary to automate the analysis ofgrids. It is also desirably to use open source software that

1http://www.neplan.ch/2http://w3.siemens.com/smartgrid/global/en/products-systems-solutions3http://www.digsilent.com

Time

Current

√2 · I ′′k

idc

ip

√2 · Ik

Fig. 1: Characteristic curve of a SC current far from agenerator

arX

iv:1

802.

0150

2v1

[cs

.CE

] 5

Feb

201

8

2

does not require license fees and allows unconstrained parallelcomputing. While many open source tools exist that allowautomation of power flow and optimal power flow evaluations,SC calculations are only supported by few tools. GenericSC calculations without considering the specifications andcorrection factors defined in the standard are supported byGridCal4, InterPSS5 and OpenDSS6. SC calculations accord-ing to IEC 60909 are considered in the open source toolElplek7, which also implements functionality for protectionsystem studies. Elplek is however a GUI application with afocus on manual single case evaluations and is neither suitedfor automated analysis nor does it scale well for large grids.There is currently no open source tool known to the authorsthat allows vectorized SC calculations according to IEC 60909including the latest revision for the consideration of DG. Thisgap is closed by the implementation described in this paper.It is available in the open source tool pandapower [8].This paper is structured as follows: Section II describes theopen source tool pandapower and how it can be used toimplement SC calculations. Section III describes how thedifferent grid elements are modelled for the fault case, andSection IV describes how the SC currents are calculated basedon the grid model. Section V shows the application of the SCcalculation in a case study from literature as well as a realMV grid. Section VI provides a summary and a conclusion ofthis paper.

II. PANDAPOWER

This paper describes the implementation of a vectorized SCcalculation considering the specifications of IEC 60909 in thegrid analysis tool pandapower.

A. About pandapower

pandapower is a BSD licensed open source moduleimplemented in Python with a data structure based on the dataanalysis library pandas [8]. It is an easy to use grid calcula-tion and analysis framework that includes equivalent circuitmodels for elements such as lines, two and three-winding-transformers, loads, external grids, synchronous generators andmore [8]. Electric elements are specified with nameplate data,such as short circuit voltages and rated apparent power fortransformers or length of lines and relative resistances in Ohm.pandapower also includes a switch model, a feature which isespecially important in radial systems. All element models arethoroughly tested and validated against commercial software.On the basis of this grid model, pandapower offers powerflow, optimal power flow, state estimation and topologicalgraph search functions. The implementation of SC calculationsaccording to IEC 60909 presented in this paper further extendsthe comprehensive grid analysis functionality of the module.

4https://github.com/SanPen/GridCal5www.interpss.com6http://smartgrid.epri.com/SimulationTool.aspx7http://pp.kpnet.fi/ijl/

(a) Bus-branch grid model

(b) Element based grid model

Fig. 2: Different grid representations

B. Grid Representation

To carry out a SC calculation, the grid parameters ofall grid elements have to be defined by the user. Thereare different possibilities of how this input parameters canbe provided. A common approach is the bus-branch model(BBM), which defines the grid as a generic collection of buseswhich are connected by branches [9] (see Figure 2a). Thebranches are modelled with a predefined equivalent circuitand can be used to model lines or transformers. However,all branch impedances and summed power injections have tobe calculated from the nameplate data of the grid elements bythe user. Instead of a generic branch model, pandapoweruses an element based model (EBM) with separate modelsfor lines, two-winding and three-winding transformers. Allelement models are internally processed with appropriateequivalent circuits to model the behaviour of the respectiveelement. Internally, all elements are combined into a BBMthat can be used for grid analysis functions. This also has theadvantage that different equivalent circuits can be specifieddepending on the grid analysis function that is being used. Forexample, pandapower provides different transformer modelsfor power flow and SC calculation. The transformer model isdefined with the same nameplate parameters by the user, butis internally modelled with a SC equivalent circuit for SCcalculations and a T-model for power flow calculations. Sincethe definition of element parameters is decoupled from thecircuit impedances, it is also possible to include correctionfactors defined for elements in the SC calculation. This wouldhave to be done manually by the user if a BBM was used.

III. SHORT CIRCUIT GRID MODEL

The grid is defined by the user with an EBM as described insection III. When a SC calculation is carried out, the EBM isconverted into a BBM. The equivalent circuit models includingspecified correction factors are considered as defined in thestandard automatically. The BBM is mathematically repre-sented in a nodal point admittance matrix Y that includes theimpedances of all grid elements [10]. This section introducesthe most important models for distribution systems analysis,all available models can be found in the pandapowerdocumentation8.

8https://pandapower.readthedocs.io

3

A. Bus Elements

There are two different kinds of bus elements which areconsidered in the short circuit calculation:• Electric motors or generators, which are modelled as a

voltage source with internal impedance• Full converter elements, which are modelled with a cur-

rent source according to the 2016 revision of the standardConstant power loads and shunt elements are neglected

according to the standard.1) Voltage Source Elements: The internal impedances of

voltage source elements can be calculated from their name-plate data. The relevant formulas are provided in the standard.The internal impedance of an external grid is calculated as [1]:

ZQ =c · UnQ√3I ′′kQ

=c · U2

nQ

S′′kQ(1)

where S′′kQ is the SC power of the grid and c is the voltagecorrection factor. Since external grid connections are equiv-alent elements representing aggregated upstream grid groupswith multiple generators, the SC power has to be retrievedfrom measurements or grid reduction. To account for worstcase situations, two values S′′kQ,min/S′′kQ,max are given aswell as minimum and maximum values for the R/X ratio ofthe grid RXQ to calculate the complex grid impedance. Thevoltage correction factor c accounts for operational deviationsfrom the nominal voltage in the grid. The standard definesdifferent correction factors which represent worst-case valuesfor minimum and maximum SC calculations as shown inTABLE I.

2) Current Source Elements: Full converter elements, e.g.PV or wind power plants, are modelled as current sources.The current injection by DG is assumed to be inductive, sothat the current is calculated as [2]:

IkU = −j(k · IrU ) (2)

where the rated current IrU and the ratio of short circuit torated current k are given by the manufacturer.

B. Branch Elements

Lines and two-winding transformers are represented by asingle SC impedance as shown in Figure 3a, and three windingtransformers are represented by a star equivalent circuit asshown in Figure 3b. Shunt admittances of all branch elementsare neglected.

1) Line: The short circuit impedance of a line is equal tothe normal operation line impedance for maximum SC currentcalculations. For minimum SC currents, the line resistance atstandard operation temperature RL20 is corrected in relationto the end temperature of the conductor θe after a fault [1]:

RL = [1 + 0.04K−1(θe − 20 ◦C)] ·RL20 (3)

TABLE I: Voltage correction factor c according to [1]

Voltage Level Tolerance cmin cmax

0.1 kV - 1 kV 6 % 0.95 1.0510 % 0.95 1.10

> 1 kV 1.00 1.10

(a) Line and Two-WindingTransformer

(b) Three-WindingTransformer

Fig. 3: Equivalent circuits for branch elements

2) Two-Winding Transformer: The relative transformerimpedance zk can be calculated from the short circuit voltageuk as:

zk =uk100·KT (4)

with the correction factor KT [1]:

KT = 0.95 · cmax

1 + 0.6 · xT(5)

where cmax is the maximum voltage correction factor on thelow voltage side of the transformer and xT is the transformerimpedance relative to the rated values of the transformer.

3) Three-Winding Transformer: Three winding transform-ers are modelled by three equivalent two-winding transform-ers [11]. The three equivalent two-winding transformers arerepresented by their SC impedance in a star connection asshown in Figure 3b. The transformer correction factor givenin equation 5 is also applied to the equivalent two-windingtransformers.

IV. SHORT-CIRCUIT CURRENT CALCULATION

The SC current is calculated in three steps:• calculate the SC contribution I ′′kI of all voltage source

elements• calculate the SC contribution I ′′kII of all current source

elements• calculate the total initial SC current I ′′k = I ′′kI + I ′′kII

This process is shown step by step for the three-bus examplegrid. The example grid is shown in Figure 4a in normaloperation. The equivalent circuit for a short circuit at Bus 3is shown in Figure 4b. The circuit is then separated into twoequivalent circuits for the calculation of I ′′kI (Figure 4c) andI ′′kII (Figure 4d). The detailed steps of how these equivalentcircuits are built and evaluated to calculate I are discussed inthis section.

A. Equivalent Voltage Source

In the first step, the voltage source contribution to the SCcurrent is calculated. The current source elements are thereforeneglected. Furthermore, all voltage sources are replaced withone equivalent voltage source at the fault location accordingto the theorem of Thevenin. The magnitude of the equivalentvoltage source is given as [1]:

UQ =c · UN√

3(6)

4

(a) 3-bus example grid

(b) Equivalent circuit for fault at bus 3

(c) Equivalent voltage source at fault location

(d) Equivalent circuit for consideration of current sources

Fig. 4: Example grid with equivalent circuits

where UN is the nominal voltage and c is either cmin for min-imum SC calculations or cmax for maximum SC calculations.Since all load currents can be neglected for SC calculations,the currents at all buses are zero except for the fault bus, wherethe current is equal to the SC current. The grid equations fora fault at bus j can then be derived from the ohmic law as:

U1...

UQj

...Un

=

Z11 . . . . . . . . . Zn1...

. . ....

... Zjj

......

. . ....

Z1n . . . . . . . . . Znn

0...

I ′′kIj...0

(7)

where the impedance matrix Z is the inverse of the nodal point

admittance matrix Y . The SC current at fault bus j can nowbe extracted from row j of the matrix equation as:

I ′′kIj =UQj

Zij

(8)

To calculate the vector of the SC currents at all buses, thematrix equation can be expanded as follows:UQ1 . . . Un1

.... . .

...

U1n . . . UQn

=

Z11 . . . Zn1

.... . .

...

Z1n . . . Znn

I ′′kI1 . . . 0

.... . .

...

0 . . . I ′′kIn

Since the current matrix is diagonal, the vector of SC currentmagnitudes at all buses can be calculated as:

I ′′kI1...

I ′′kIn

=

UQ1

Z11

...UQn

Znn

(9)

B. Contribution of Current Source Elements

To calculate the current source component of the SC current,all voltage sources are short circuited and only current sourcesare considered. The bus currents are then given as:

I1...Im

...In

=

0

...I ′′kIIj

...0

−

I ′′kC1

...I ′′kCj

...I ′′kCn

=

−I ′′kC1

...I ′′kIIj − I

′′kCj

...−I ′′kCn

(10)

where I ′′kC are the SC currents that are fed in by the converterelement at each bus and I ′′IIj is the contribution of converterelements at the fault bus j. With the voltage at the fault busknown to be zero, the network equations are given as:

U1...

0

...Un

=

Z11 . . . . . . . . . Zn1...

. . ....

... Zjj

......

. . ....

Z1n . . . . . . . . . Znn

−I ′′kC1

...I ′′kIIj − I

′′kCj

...−I ′′kCn

(11)

From which row j of the equation yields:

0 = Zjj · I′′kIIj −

n∑m=1

Zjm · IkCj (12)

which can be converted into:

I ′′kIIj =1

Zjj

·n∑

m=1

Zjm · IkC,m (13)

5

(a) HV example network with wind parks

Bus 1 Bus 2 Bus 3 Bus 40

5

10

15

20

25

I′′ k [k

A]

20.00

2.9133.707

2.399

21.26

3.9034.795

3.215

Without Wind ParksWith Wind Parks

(b) pandapower results for validation of example HV network

Fig. 5: Validation of SC implementation with three wind parksexample from [2]

To calculate all SC currents for faults at each bus simul-taneously, this can be generalized into the following matrixequation:

I ′′kII1

...

...

I ′′kIIn

=

Z11 . . . . . . Zn1

.... . .

......

. . ....

Z1n . . . . . . Znn

I′′kC1

Z11

...

...I′′kCn

Znn

(14)

C. Example Network

For the example network shown in Figure 4a, a calculationof the maximum initial SC current with a voltage tolerance of10% yields the following result:

I ′′k =

145.0733.8444.524

kA+

0.1810.2080.117

kA =

145.2544.0524.641

kAThe detailed derivation of these currents including the defini-tion of the nodal point admittance matrix is given in AppendixA.

V. EXAMPLES AND CASE STUDIES

This section includes some exemplary case studies todemonstrate how the pandapower SC module can be used.

Secondary SubstationSwitching StationPrimary SubstationLineTie-Line

Fig. 6: Real MV example grid

A. Impact of Wind Parks in a High Voltage Network

The example from [2] is used to validate the implementationof the SC calculation with full converter elements. Accordingto [2], a fault at Bus 2 in the network depicted in Figure 5awith k = 1.2 leads to a SC current of:

I ′′k2 = 2.913 kA + 0.990 kA = 3.903 kA

To validate the implementation presented in this paper, aSC calculation with and without wind parks is run withpandapower. The results can be seen in Figure 5b andare consistent with the results given in [2]. Because of thevectorized implementation, pandapower also returns the SCcurrents at all other buses.

B. MV Network Case Study

The short circuit calculation is applied in a case studyto the real Medium Voltage (MV) distribution grid shownin Figure 6. The distribution grid is operated at a nominalvoltage of 20 kV and is connected to the 110 kV high voltage(HV) network through two HV/MV substations. It services282 MV/LV stations with a combined peak load of about11 MW. In addition, there are DG units with a rated powerof over 13 MW installed in this grid area. With this highpenetration of DG, this area is well suited to demonstratethe impact of DG on the short circuit current in a real usecase. Figure 7 shows the maximum (Figure 7a) and minimum(Figure 7b) short circuit currents for faults at all buses. Sincethe network is operated radially, the impedance between faultand external grid connection increases the larger the distancebetween fault location and HV/MV substation. That is whythe short circuit current decreases with growing distance to theHV/MV substation. The overall maximum short circuit currentcan therefore be found directly at the HV/MV substation,whereas the minimal short circuit current is found at the endof the feeder. The short circuit contribution of the DG unitsis considered with k = 1.0. When considering DG in theshort circuit calculation according to the 2016 revision of thestandard, the maximum short circuit current at the primarysubstation rises from 6.65 kA to 7.69 kA (see Figure 7a). Thisrepresents an increase by 16 %, which shows that consideringthe short circuit contribution of DG can be relevant when theshort-circuit capabilities of bus bars in the primary substations

6

0 2 4 6 8 10 12Distance from Substation

2

3

4

5

6

7

8S

hort

Circ

uit C

urre

nt [k

A]

Without DG unitsWith DG units

(a) Maximum Initial Short Circuit Currents at all buses

0 2 4 6 8 10 12Distance from Substation

1

2

3

4

5

6

7

8

Sho

rt C

ircui

t Cur

rent

[kA

]

Without DG unitsWith DG units

(b) Minimum Initial Short Circuit Currents at all buses

Fig. 7: Real MV example grid

are assessed. It can also be seen that the short circuit currentis the same at the beginning of all feeders when DG arenot considered, since the short circuit current is fed onlyby the external grid connection. When DG are considered,the individual feeders show a different behaviour dependingon the DG capacity on each feeder, which makes a detailedanalysis of the feeders necessary. For the minimum shortcircuit calculation, the end temperature of the lines after thefault is assumed to be 200 ◦C for overhead lines and 90 ◦C forunderground cables. The minimum short circuit current at theend of the feeder then rises by 6 % from 1.99 kA to 2.11 kAwhen considering DG (see Figure 7b). While maximum SCcurrents are relevant for the SC current capability of thegrid components, minimum SC currents are relevant whenanalysing if all faults can be reliably detected by protectionsystems.

VI. CONCLUSION

This paper presents a method for the calculation of the initialshort circuit current according to IEC 60909. Full converterelements are considered as current sources according the 2016revision of the standard. The presented method is implementedin the network analysis tool pandapower, which is availableas an open source software. The implementation allows totake lines, two-winding transformers, three-winding transform-ers, synchronous generators, asynchronous machines and fullconverter elements into account. While calculation of othershort circuit currents like the peak current ip or the thermalequivalent current Ith are not addressed in this paper, they arealso available in the pandapower implementation. Resultsfor all elements and short circuit currents are tested and val-idated against commercial software as well as against resultsfrom other publications. The vectorized implementation allowsefficient simultaneous calculation of SC currents at all buses

even for large networks. This is especially useful in networkplanning applications, where multiple possible fault scenarioshave to be taken into account. Since pandapower is focusedon automated evaluations, the presented implementation iswell suited to consider SC currents in automated networkanalysis and optimization studies. The practical applicabilityof the implementation was demonstrated with a case study ofa real MV grid with a high penetration of DG. The case studyshowed, that DG can have a significant impact on the shortcircuit currents in distribution systems.

The implementation presented in this paper is the first com-prehensive implementation of the IEC 60909 standard in anyopen source power systems analysis tool. It is therefore a valu-able contribution towards closing the gap between commercialand open source power system analysis tools. pandapower isunder continuous development on github [12], and it is plannedto add further functionality to the short circuit calculation, suchas short circuit currents close to generators or single-phasefaults.

ACKNOWLEDGEMENT

This work was supported by the German Federal Ministryfor Economic Affairs and Energy and the Projekttrager JulichGmbH (PTJ) within the framework of the projects Smart GridModels (FKZ: 0325616). The authors furthermore thank NetzeBW GmbH for providing grid data as well as helpful feedbackand discussions on the topic.

REFERENCES

[1] “IEC 60909-0:2016: Short-circuit currents in three-phase a.c. systems -part 0: Calculation of currents,” International Standard, 2016.

[2] G. Balzer, “Short-circuit calculation with fullsize converters accordingto iec 60909,” in 21st Conference of Electric Power Supply Industry,CEPSI, Bangkok, Thailand, October 2016.

[3] T. Nippert, T. Behrens, and K. H. Kny, “Short-circuit current calculationaccording to din vde 0102 in planning and operation of an urbandistribution system,” in CIRED 2005 - 18th International Conferenceand Exhibition on Electricity Distribution, June 2005, pp. 1–6.

[4] D. Sweeting, “Applying iec 60909, fault current calculations,” IEEETransactions on Industry Applications, vol. 48, no. 2, pp. 575–580,March 2012.

[5] P. Karaliolios, A. Ishchenko, E. Coster, J. Myrzik, and W. Kling,“Overview of short-circuit contribution of various distributed generatorson the distribution network,” in 2008 43rd International UniversitiesPower Engineering Conference, Sept 2008, pp. 1–6.

[6] S. Boljevic and M. F. Conlon, “Fault current level issues for urbandistribution network with high penetration of distributed generation,”in 2009 6th International Conference on the European Energy Market,May 2009, pp. 1–6.

[7] K. Tanaka and K. Takahashi, “An efficient method of modifying z-matrixelements in short-circuit capacity calculations,” Electrical Engineeringin Japan, vol. 114, no. 2, pp. 48–56, 1994.

[8] L. Thurner, A. Scheidler, Schafer et al. (2017) pandapower - anOpen Source Python Tool for Convenient Modeling, Analysis andOptimization of Electric Power Systems. Preprint. [Online]. Available:https://arxiv.org/abs/1709.06743

[9] R. Zimmerman, C. Murillo-Sanchez, and R. Thomas, “Matpower:Steady-state operations, planning, and analysis tools for power systemsresearch and education,” Power Systems, IEEE Transactions on, vol. 26,no. 1, pp. 12–19, Feb 2011.

[10] J. J. Grainger and W. D. Stevenson, Power system analysis. McGraw-Hill, 1994.

[11] “IEEE recommended practice for industrial and commercial powersystems analysis,” IEEE Std 399-1997, pp. 1–488, Aug 1998.

[12] pandapower github repository. [Online]. Available: https://github.com/lthurner/pandapower